Abstract
Chomsky (1988) has suggested that human language depends on the evolution of a unique computational capacity that makes possible recursively constructed symbolic representations of the experienced world. I review evidence from the animal cognition literature suggesting that the construction of abstract representations that support complex inferences is widespread. It is found even in insects. This widespread symbol processing capacity supports the construction of complex data structures (symbolic propositions). In vertebrates at least, there is evidence that the representation of actions takes the predicate-argument form characteristic of human language. Thus, the symbol processing capacity underlying the thoughts expressed in human language appears to be evolutionarily ancient. What is unique to humans is the ability to translate these private representations into a communicable symbol system of comparable representational power.
ACKNOWLEDGMENTS
The author is grateful for helpful comments and suggestions by Randolf Menzel, Lila Gleitman, and this journal's editors, and for research support from NIMH RO1MH077027.